File size: 3,150 Bytes
2abd452 | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 | """Verify released generated rows against their untouched raw batch tables."""
import hashlib
import json
from pathlib import Path
import pandas as pd
def validate():
release = Path(__file__).resolve().parents[1]
root = release / "generated/scddpm_rfr_fixed_range"
rows = pd.read_csv(root / "Generated_Designs.csv.gz", float_precision="round_trip")
manifest_path = release / "file_manifest.json"
manifest = json.loads(manifest_path.read_text())
# Validate the release's original raw files against the recorded hashes.
hashed = 0
for record in manifest:
if not record["path"].startswith("generated/scddpm_rfr_fixed_range/raw/"):
continue
with (release / record["path"]).open("rb") as stream:
observed = hashlib.file_digest(stream, "sha256").hexdigest()
if observed != record["sha256"]:
raise RuntimeError(f"Released raw file hash differs: {record['path']}")
hashed += 1
# Compare all decoded inputs and all saved outcome fields, not merely the
# aggregate totals. Original evaluator indices are intentionally reused.
compared = 0
for voyage_range in (1000, 5000, 8000):
for batch in range(10):
batch_root = root / f"raw/range_{voyage_range}/batch_{batch:02d}"
released = rows.loc[rows.Range_nmi.eq(voyage_range) & rows.Batch.eq(batch)].sort_values("Sample_In_Batch")
for filename in ("Submitted_Designs_189.csv.gz", "Exact_Results.csv.gz"):
original = pd.read_csv(batch_root / filename, float_precision="round_trip").sort_values("Sample_In_Batch")
pd.testing.assert_frame_equal(
released.loc[:, original.columns].reset_index(drop=True),
original.reset_index(drop=True), check_dtype=False, check_exact=True,
)
compared += len(released)
if compared != 3000 or rows.Sample_ID.duplicated().any():
raise RuntimeError("The release lost or duplicated generated attempt identities")
report = {"attempts_compared": compared, "raw_file_hashes_verified": hashed,
"all_decoded_inputs_match_raw": True, "all_saved_outcomes_match_raw": True,
"duplicate_sample_ids": int(rows.Sample_ID.duplicated().sum()),
"completed": int(rows.Evaluator_Completed.eq(True).sum()),
"feasible": int((rows.Evaluator_Completed.eq(True) & rows.Global_V1p1_Feasible.eq(True)).sum()),
"evaluation_failures_retained": int(rows.Evaluator_Completed.eq(False).sum())}
output = root / "provenance/Validation.json"
output.write_text(json.dumps(report, indent=2) + "\n")
relative = output.relative_to(release).as_posix()
manifest = [row for row in manifest if row["path"] != relative]
manifest.append({"path": relative, "bytes": output.stat().st_size,
"sha256": hashlib.sha256(output.read_bytes()).hexdigest()})
manifest_path.write_text(json.dumps(sorted(manifest, key=lambda row: row["path"]), indent=2) + "\n")
print(json.dumps(report, indent=2))
if __name__ == "__main__":
validate()
|